Motion Analysis: Past, Present and Future

The subject of motion has been the center of interdisciplinary studies since the time when Zeno posed his paradox circa 500BC. However, computer vision, the use of a camera and a computer to recognize objects, people and/or events automatically, is a relatively young field of research. Its development began in the early 1960s; however, it has matured fairly quickly. Today, it is contributing to the solutions of some of the most serious societal problems. Motion analysis of a sequence of images is an important part of computer vision. This chapter briefly presents the contributions to motion analysis from other fields followed by the computer vision-based analysis of motion from a sequence of images. Analysis and understanding of images based on both feature tracking and optical flow estimation are presented. Early works focused on the computation of structure from motion of objects from a sequence of images via point features. This was followed by the computation of optical flow to characterize motion. Applications today focus on the monitoring of traffic, providing guidance to a motorist in terms of his/her position relative to traffic lanes and traffic ahead, and inspection of complicated three-dimensional industrial parts, to mention a few. Research focus has shifted from inanimate objects to people, for example monitoring people and their activities in public places or monitoring activities from an unmanned aerial vehicle. These applications are dominating the research scene through the belief that computer vision/motion analysis can contribute to the solution of societal surveillance and biometric problems. The chapter ends with a discussion of the future directions of research in motion analysis and possible applications.